[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Qemu-block] [PATCH v2] mirror: double performance of the bulk stage

From: Denis V. Lunev
Subject: Re: [Qemu-block] [PATCH v2] mirror: double performance of the bulk stage if the disc is full
Date: Wed, 20 Jul 2016 23:30:39 +0300
User-agent: Mozilla/5.0 (X11; Linux x86_64; rv:38.0) Gecko/20100101 Thunderbird/38.5.1

On 07/20/2016 10:08 PM, Paolo Bonzini wrote:

----- Original Message -----
From: "Denis V. Lunev" <address@hidden>
To: "Vladimir Sementsov-Ogievskiy" <address@hidden>, address@hidden, 
Cc: address@hidden, address@hidden, address@hidden, address@hidden, 
address@hidden, "Kevin Wolf" <address@hidden>
Sent: Wednesday, July 20, 2016 7:36:00 PM
Subject: Re: [PATCH v2] mirror: double performance of the bulk stage if the 
disc is full

On 07/14/2016 08:19 PM, Vladimir Sementsov-Ogievskiy wrote:
Mirror can do up to 16 in-flight requests, but actually on full copy
(the whole source disk is non-zero) in-flight is always 1. This happens
as the request is not limited in size: the data occupies maximum available
capacity of s->buf.

The patch limits the size of the request to some artificial constant
(1 Mb here), which is not that big or small. This effectively enables
back parallelism in mirror code as it was designed.

The result is important: the time to migrate 10 Gb disk is reduced from
~350 sec to 170 sec.

Signed-off-by: Vladimir Sementsov-Ogievskiy <address@hidden>
Signed-off-by: Denis V. Lunev <address@hidden>
CC: Stefan Hajnoczi <address@hidden>
CC: Fam Zheng <address@hidden>
CC: Kevin Wolf <address@hidden>
CC: Max Reitz <address@hidden>
CC: Jeff Cody <address@hidden>
CC: Eric Blake <address@hidden>

v2: in case of s->buf_size larger than default use it to limit io_sectors

   block/mirror.c | 10 ++++++++--
   1 file changed, 8 insertions(+), 2 deletions(-)

diff --git a/block/mirror.c b/block/mirror.c
index b1e633e..3ac3b4d 100644
--- a/block/mirror.c
+++ b/block/mirror.c
@@ -23,7 +23,9 @@
#define SLICE_TIME 100000000ULL /* ns */
   #define MAX_IN_FLIGHT 16
-#define DEFAULT_MIRROR_BUF_SIZE   (10 << 20)
+#define MAX_IO_SECTORS ((1 << 20) >> BDRV_SECTOR_BITS) /* 1 Mb */
/* The mirroring buffer is a list of granularity-sized chunks.
    * Free chunks are organized in a list.
@@ -322,6 +324,8 @@ static uint64_t coroutine_fn
mirror_iteration(MirrorBlockJob *s)
       int nb_chunks = 1;
       int64_t end = s->bdev_length / BDRV_SECTOR_SIZE;
       int sectors_per_chunk = s->granularity >> BDRV_SECTOR_BITS;
+    int max_io_sectors = MAX((s->buf_size >> BDRV_SECTOR_BITS) /
+                             MAX_IO_SECTORS);
sector_num = hbitmap_iter_next(&s->hbi);
       if (sector_num < 0) {
@@ -385,7 +389,9 @@ static uint64_t coroutine_fn
mirror_iteration(MirrorBlockJob *s)
                                             nb_chunks * sectors_per_chunk,
                                             &io_sectors, &file);
           if (ret < 0) {
-            io_sectors = nb_chunks * sectors_per_chunk;
+            io_sectors = MIN(nb_chunks * sectors_per_chunk,
+        } else if (ret & BDRV_BLOCK_DATA) {
+            io_sectors = MIN(io_sectors, max_io_sectors);
io_sectors -= io_sectors % sectors_per_chunk;
guys, what about this patch?
I think that at this point it has missed hard freeze.  It's not up to me
whether to consider it a bugfix.

I see. It's pity this patch has missed deadline...
The difference is really big.

OK ;) I think it will not be missed in the next


reply via email to

[Prev in Thread] Current Thread [Next in Thread]